New Challenges to Arms Export Control: Whither Wassenaar?

RON SMITH & BERNARD UDIS

RON SMITH & BERNARD UDIS1

Ron Smith is Professor of Economics, Birkbeck College, University of London. He has been Visiting Professor at the London Business School and the University of Colorado and has published seven books and over a hundred papers in the areas of applied econometrics and defense economics. Bernard Udis is Professor of Economics, Emeritus, at the University of Colorado at Boulder. During 1997, he was a Foster Fellow at the U.S. Arms Control and Disarmament Agency, where his responsibilities included the Wassenaar Arrangement. In addition to several books on the adjustment of the U.S. and European economies to reduced military spending, his publica- tions on defense economics have appeared in numerous scholarly journals.

t is widely recognized that arms exports can have nega- Use Goods and Technologies (WA). The end of the Cold tive externalities on national security; as a result, there War allowed the formation of the WA, the first multilat- Iare almost universal national controls on this indus- eral agreement covering both conventional weapons and try. There are also a variety of international organizations sensitive dual-use goods and technologies. The member that attempt to regulate the international trade in arms list includes Russia, but not China, and is designed to pre- through the coordination of suppliers. Paul Levine and vent destabilizing acquisitions of weapons and technolo- Ron Smith developed a theory of such coordination that gies through a formal process of transparency and examines the interaction that occurs in the coordination consultation. The WA is the successor to the now defunct of the supply of weapons through international export Coordinating Committee for Multilateral Export Controls controls, the coordination of the production of weapons (COCOM), which was created and designed to restrict through collaboration, and the coordination of the levels the transfer of arms and dual-use technologies to com- of military expenditure through alliances.2 In this article, munist countries during the Cold War.3 Andrew Pierre we wish to examine the institutional features of arms ex- referred to the WA as “…[t]he best hope for creating a port controls: the practical and logistic issues involved, the supplier-based multilateral regime.”4 William Keller and international organizations that are working to regulate ex- Janne Nolan noted, however, that the WA “…has received ports, and the factors that determine the effectiveness of scant attention from the policy community and ridicule the various arrangements. from the arms lobby, and it is presently languishing with 5 Of the large number of export control organizations that no high level involvement.” exist, we have chosen to focus on the Wassenaar Arrange- In surveying the challenges, we first set out the back- ment on Export Controls for Conventional Arms and Dual- ground of what is currently controlled, why these particular

The Nonproliferation Review/Summer 2001 81 RON SMITH & BERNARD UDIS

items have been chosen, and how the controls are insti- Within conventional weapons, it is useful to distinguish tuted. Second, we discuss some general theoretical issues major systems from light weapons. The latter include small that arise in international agreements of this sort and, in arms, land mines, small mortars, and man-portable mis- particular, the ways in which arms export controls differ siles. While light weapons have probably caused 90 per- from classic arms control. Third, we address the disrup- cent of the casualties in recent wars,9 major weapons tive potential of technical change. Finally, in light of this systems have the potential to change regional balances of analysis, we then return to examine the role of the WA. power. Light weapons are also more difficult to control Many of these issues have been discussed extensively in than major systems, partly because whereas major sys- the literature on this topic, which we have summarized tems are supplied by an oligopoly, light weapons are com- briefly.6 petitively supplied by large numbers of producers.10 With the exception of the anti-personnel landmine agreement, CONTROLLING THE INTERNATIONAL FLOW there has been little action on light weapon control.11 Dual- OF ARMS use systems raise unique difficulties, partly because many of the technologies required to produce WMD and their Although they overlap, it is useful to distinguish five delivery systems have important industrial applications main categories of militarily useful goods and services that (i.e., nuclear, biological, chemical, and space) and partly may be subject to control. These are: weapons of mass because, as COCOM found, it is difficult to define mili- destruction (WMD), major weapons systems, light weap- tarily-relevant technologies. Services are not generally sub- ons, dual-use technologies, and services (e.g., training and ject to control, but are increasingly important in the transfer maintenance). Each category raises very different issues of military knowledge and technology. Without training and involves very different market structures and control and support services, many states have difficulty using mechanisms. There is a widespread, though far from the advanced weapons they acquire. unanimous, consensus that the proliferation of WMD— including chemical, biological, radiological, and nuclear There are a variety of reasons for which states attempt (CBRN)—and their delivery systems should be controlled. to control arms exports. The simplest is strategic embargo: This trend is reflected in a set of treaties and supplier to prevent the sale of weapons to a potential enemy to groups that have been created and instituted, discussed limit its military capability. This was the main objective below. Whatever their limitations, illustrated by the 1998 of COCOM. As Michael Mastanduno has pointed out, Indian and Pakistani nuclear tests, these arrangements have however, a recurrent source of conflict between the United nonetheless inhibited the proliferation of WMD. In con- States and its allies was whether the objective of COCOM trast, there is no such consensus that there should be gen- was strategic embargo or more general economic war- eral controls on the sales of conventional weapons, since fare.12 Strategic embargos may overlap with the preven- a state’s right to self defense, embodied in the U.N. Char- tion of the proliferation of particular types of arms, such ter, gives it the right to buy arms from abroad. as inhumane weapons or WMD, either in general or to particular pariah states like North Korea, Iraq, Iran, Libya, However, there may be specific reasons to control par- and previously South Africa. Suppliers may also fear in ticular transfers, and the 1991 declaration by the five per- some cases that the external supply of weapons may pro- manent members of the U.N. Security Council (P5) long a war—hence the embargo of military supplies to outlines such reasons.7 Typically, transfers should be re- the former Yugoslavia. In yet another situation, states may stricted if the recipient’s acquisition of arms is excessive fear that the introduction of particular arms in a region, or destabilizing. Not only is it difficult to define what a such as high technology weapons, may destabilize inter- “destabilizing acquisition” of weapons entails, but some, national relations, either through an expensive arms race like Colin Gray, argue that the concept itself is incoher- or through encouraging pre-emptive aggression. For in- ent: a category mistake, since it is governments, not weap- stance, until late 1997, U.S. policy did not permit selling ons, that cause war.8 Nonetheless, various export control F-16s to Latin America (the exception was to Venezuela, arrangements have attempted and are attempting to de- where there were particular U.S. security concerns regard- fine the notion of a “destabilizing acquisition,” either in ing Cuba). Supplier states may prohibit arms sales to gov- terms of the capabilities of the weapons or the character- ernments with poor human rights records, either as a istics of the recipients. general indication of disapproval or with the concern that

82 The Nonproliferation Review/Summer 2001 RON SMITH & BERNARD UDIS weapons may be used in internal repression. The embargo amples include the Nuclear Non-Proliferation Treaty on China after the Tiananmen Square incident, as well as (NPT), the Landmine Treaty, the Chemical Weapons Con- objections to selling to Indonesia during its illegal occupa- vention (CWC), and the Biological and Toxin Weapons tion of East Timor, are examples. Finally, suppliers may Convention (BTWC).16 These treaties are open to all be concerned that the cost of the arms purchases could states, and a large number of countries typically sign on damage the recipient’s economy, through the excessive (often over 90). By joining and becoming a member of accumulation of debt or the diversion of resources from “the club,” states gain international reputation benefits (and development needs. sometimes direct economic benefits, as with the NPT). These objectives are wider and more vague than those For most, the costs of signing are not high. A majority of of classic arms control, which theoretically aims to: (1) the signatories do not seek these weapons in the first place, reduce the probability of war; (2) reduce the adverse con- and, for those that do, enforcement is not very strict and sequences of war should it occur; and (3) reduce the cost punishment for violation rare. For instance, the 1972 of military preparations.13 Frequently, there are trade-offs BTWC has no mechanism to check compliance, though between these objectives: for example, measures that re- discussion about the development of a verification proto- duce the probability of war could, in effect, increase costs. col to the treaty has been ongoing since 1994. There also However, the trade-offs in traditional arms control are exists a range of transparency measures, of which the U.N. much simpler than those in arms export control. For in- arms transfer register is the most important. The United stance, if arms control prohibits a state from acquiring a Nations requests states to report transfers of seven types system, such as an anti-ballistic missile system (ABM), it of arms: battle tanks, armored combat vehicles, large cali- saves on expenses; however, if export controls stop a state ber artillery systems, combat aircraft, attack helicopters, from making a sale to another country, money is lost.14 warships, and missiles/missile launchers. Over 90 coun- We return to the differences between traditional arms con- tries have reported transfers. The WA employs these same trol and arms export control below. categories, and the first five were used in the Conven- tional Forces in Europe (CFE) Treaty. There are a variety of international mechanisms for controlling arms exports, but most rely on national export Finally, in addition to embargoes/sanctions, treaties, and regulations. These usually involve: lists of products that transparency measures, there is a patchwork quilt of sup- 17 require a license/notification; lists of countries to which plier agreements, sometimes referred to as regimes. exports can or cannot be made; a list of criteria used to These more restrictive “clubs” include: the Nuclear Sup- judge uncertain cases; customs procedures to stop unli- pliers Group (NSG), established in 1975 as the “London censed exports; a system to guarantee the end use of the Club”; the Missile Technology Control Regime (MTCR), equipment and prohibit third-party transfers; and, in mul- established in 1987; the Australia Group (AG), formed in tilateral systems, a notification mechanism for reporting 1985 to control chemical and biological weapons (CBW); exports that are denied, as well as perhaps a “no-under- and the Wassenaar Arrangement, established in 1996 to cutting rule” (i.e., states agree to deny sales that are re- monitor the transfer of conventional weaponry. Often, both fused by another member state). Not all countries possess treaties and supplier groups cover the same categories of well-functioning export control systems and, as the Iraq equipment, materiel, and technology; but, whereas the case demonstrated, even reasonably well-functioning sys- former are formal and open to all, the latter tend to be tems can be evaded. The developments in the European more informal with restricted membership. In general, Union (EU) system of arms export controls should be seen supplier groups originated in loose cooperation among a as an extension of national export controls, since the single small number of countries. Most have become organized market, which removes restrictions on cross-border trade more formally and expanded to typically about 30 indus- within the EU, has reduced the effectiveness of national trialized states. Non-member states may choose to adhere controls.15 to the guidelines of these arrangements without being members. In contrast to today’s multilateral emphasis, There is a range of U.N. embargoes or sanctions that COCOM was a supplier group of U.S. allies, and the con- prohibit the supply of weapons to particular states. There trols were aimed at the rival Soviet bloc. The neutral states, is also an array of treaties, by which participants agree to though not members, tended to abide by COCOM con- restrict their acquisition of certain sorts of weapons. Ex- trols broadly, albeit often after U.S. pressure was applied.

The Nonproliferation Review/Summer 2001 83 RON SMITH & BERNARD UDIS

Again, in the case of supplier groups, there may be eco- interest of the agent ex post to carry out the threat; and nomic incentives to join; for example, the promise of space (5) the sequence of moves (e.g., whether there is a leader cooperation with the United States encouraged Russia to or not). The credibility of the threat to punish defectors comply with MTCR restrictions.18 Because these are ar- can be increased through various forms of precommitment; rangements or agreements rather than treaties, they are for example, ceasing certain activities or severing particu- enforced, like COCOM, by the national legislation of lar relationships can be useful in establishing legitimacy. members. Notice that in this context it is usual to think of Leadership plays a central role in creating a coalition and suppliers as states, whereas it is companies, often in reaching consensus on particular issues, defined during transnational, that do the actual supplying.19 the negotiations. In the case of export controls, leader- The natural supplier group—small, dominating a large ship may be provided by small countries that act as facili- part of the market, and with no ambiguities about mem- tators; yet, it is more often provided by large countries, bership—is the P5. These five states began discussions particularly the United States. and drew up a set of guidelines; but after the decision by One general question that arises in such games con- President George Bush Sr. to sell F-16s to Taiwan, China cerns the performance of a centralized organization ver- ceased cooperation and has not participated since. Direct sus decentralized trigger strategies, which punish defection. export controls are not the only way to restrict arms trans- The most famous of these trigger strategies is the “tit-for- fers. The United States bought 21 Moldovan Mig-29 fight- tat” strategy embodied in the concept of the prisoner’s ers under the Department of Defense (DOD) Cooperative dilemma. One begins by cooperating in the first move and Threat Reduction (Nunn-Lugar) program, to prevent them then continues by mimicking the first corresponding move from being acquired by Iran.20 This program was also used of the opponent (or cooperator, as the case may be).24 to help persuade Belarus, Ukraine, and Kazakhstan to re- The main problem with “tit-for-tat” is that it is vulnerable linquish Soviet-origin nuclear weapons on their territory.21 to imperfect monitoring of whether the other agents are cooperating or defecting.25 In general, trigger strategies do THEORETICAL ISSUES OF ARMS EXPORT not necessarily enforce the first best cooperative outcome. CONTROL In addition, states cannot rely on this type of interaction, due to its inherent unpredictability. If arms exporters can agree to coordinate their actions, the impact of controls is increased. Thus, collective ac- Instead, states have chosen to create international or- tion and cooperation are more effective in meeting com- ganizations to try and solve the collective action problem. mon arms export control objectives than individual action. The reasons why states choose to act through interna- However, individual states have an incentive to defect from tional organizations are explored by Kenneth Abbott and the agreement, as the short-term economic incentives to Duncan Snidal, who noted that international organizations sell may be more attractive than the long-term security provide centralization (a concrete and stable administra- benefits.22 Therefore, export controls contain both “pub- tive apparatus managing collective activities) and indepen- lic good” (i.e., non-rivalry, non-excludability) as well as dence (the authority to act with a degree of autonomy prisoner’s dilemma “cartel stability” components (i.e., there and often with neutrality in defined spheres).26 This au- is a joint interest to restrict supply, but an individual inter- tonomy allows organizations to “launder” activities, which est to cheat if detection can be avoided). These are re- would be unacceptable if performed by a single state. In- peated games; individual suppliers in the group decide ternational organizations also reduce transaction costs, whether to abide by the agreement or to defect over a internalize externalities, gain economies of scale, and pool sequence of possible sales.23 costs and risks. They can act as trustee, allocator, information provider, and arbiter. Certainly arms export con- While the game theory literature has not produced ro- trol organizations help share intelligence and operating bust predictions, it does suggest that five factors affect procedures, usually through a flow from large to small the outcome of such games: (1) the structure of payoffs; member countries.27 However, supplier arrangements like (2) the discount rate or myopia of the agents (i.e., how the WA have almost no autonomy compared to interna- they weight long- and short-term factors); (3) how well tional organizations like the North Atlantic Treaty Orga- agents can monitor each other’s behavior; (4) the cred- nization (NATO) or the International Monetary Fund ibility of threats, in particular whether it would be in the

84 The Nonproliferation Review/Summer 2001 RON SMITH & BERNARD UDIS

(IMF). In the case of export controls, states have so far Although they now appear to overlap, the origins of been unwilling to delegate authority to a centralized agent. arms control and arms export control are quite distinct. Given that there is a strong case for developing an in- Arms control originally involved agreements between en- ternational organization to oversee arms export controls, emies, whereas arms export control involved agreements the obvious question remains: why are there so many sup- between allies to reduce the military capability of enemies plier groups performing such closely related tasks? Why through embargo. Overlap occurs because nominal allies is there a patchwork quilt rather than an overarching or- in arms export control (the United States, Russia, and ganization?28 A possible analogy to explain this phenom- France in the WA, for instance) behave like enemies in enon involves numerical problems containing many linked arms control agreements. variables and conflicting constraints. In such cases, good Classic theories of arms control ask the following: what solutions to complicated conflict-laden calculations can incentives exist for states to come to an agreement, and best be found by breaking the entire problem into non- what incentives exist for the states subject to the agree- overlapping domains, patches, with some coupling be- ment to comply? These incentives are, in fact, a function tween the patches.29 Competing organizations that develop of the technologies involved, institutional structure, and unique approaches can share information, so that solu- the verification possibilities. For classic arms control to tions or practices that work at one can be transmitted to work, all parties to the agreement must benefit. One must the others, as weapon lists were transferred from the CFE be able to draw up a contract that can be monitored and Treaty to the U.N. Register, to the WA. Pierre empha- verified, and there must be penalties for non-compliance. sizes the importance of a composite approach: “…a mul- Arms control works easiest where there are a small num- tifaceted regime with overlapping and complementary ber of parties involved (easier to negotiate and more likely institutions and initiatives, none of which may be fully to benefit all parties); there is an identifiable object (piece adequate by themselves but which are mutually reinforc- of equipment or technology that can be monitored); that ing.”30 There are two dimensions to the optimal size of object has an unambiguous military purpose; and control an arms export control patch: (1) the range of activities of that object restricts destabilizing military capability. Even that it controls; and (2) the range of members that it in- in the classic framework, there are problems of distinguish- cludes. In principle, each margin would be determined by ing between avoidance and evasion, and between what a comparison of costs and benefits of extension. The the parties know and what they can prove. Arms export marginal member or activity would be the one at which control, in contrast, tends to involve fairly large numbers the costs of extension just equaled the benefits. of parties, diffuse incentives and punishments, severe moni- It is worth returning to the distinction between classic toring and punishment problems, and poorly identified arms control and arms export controls.31 Todd Sandler objects of control that make verification difficult. The and Keith Hartley emphasized two features of classic arms problems that COCOM faced were illustrative in each of control: these respects. • First, the greater its success in achieving security and The classic arms control framework does not apply to stability, the less important it becomes. The 1817 agree- arms export control, in that the targets of control—poten- ment between the United States and Britain to limit naval tial recipients—are not willing parties to the contract. They vessels on the Great Lakes and Lake Champlain is an are either excluded from agreements, which are between example; compliance with this agreement has not been suppliers, or coerced to agree, as was the case with Iraq an issue in U.S.-U.K. relations. at the end of the Gulf War. From the demand side, export • Second, arms control is only needed in an adversarial controls appear to be hypocritical, selective, and discrimi- environment. However, in such an environment the natory.34 Recipients may recognize the joint interest of level of trust and confidence required to negotiate an supplier groups in restricting the flow of destabilizing weap- agreement rarely exists. ons, but they are more likely to accuse suppliers of form- A paradox remains: when achievable, arms control is ing “cartels” to restrict the flow of goods or technology in not needed; and when needed, it is not achievable.32 Arms order to raise prices and maintain military dominance and control exists on this fine line between necessity and joint monopoly. Countries, like firms, must choose how achievability. It exists on the edge of chaos, a characteris- to source essential inputs; whether by buying on the open tic of many adaptive systems.33 market, within established relationships, or through self-

The Nonproliferation Review/Summer 2001 85 RON SMITH & BERNARD UDIS

sufficiency. Restricting supply to recipients alters those cremental. For example, it may be necessary to first choices. Supply restrictions make it more attractive for establish procedures and precedents in relatively states to set up their own indigenous arms industry, to uncontroversial areas, such as inhumane weapons, and develop unconventional weapons, and to use unorthodox then continue to extend the scope as confidence in the methods of transfer, all of which reduce the transparency arrangement grows. If such organizations are effective, of arms acquisition.35 Because export control involves the they may assist in the establishment of a set of norms regulation of others rather than cooperation, it tends to that maintain compliance.38 Existing arms export control reflect the classic forms of the regulatory dialectic char- organizations are a long way from establishing strong acteristic of financial markets. Regulations are avoided and norms of this sort. evaded, and then new regulations are required.36 Any control system involves compliance costs, and the TECHNOLOGY ISSUES costs of export controls fall heavily on commercial firms, The impact of technology on arms control is not a new which are the potential suppliers. At some stage, compli- problem. The 1922 Washington Naval Treaty restricted ance costs may outweigh the regulatory benefits for the the range of Japanese battleships, giving Japan incentives supplying countries. However, since it is the firms that to circumvent the constraint through the development of pay the cost and the government/military that reaps the aircraft carriers.39 COCOM, while not watertight, did raise benefit, states must make a distributional judgment. In the the cost of gaining technology and weapons for the So- United States, this judgment is sometimes expressed in viet Union, but there were problems addressing the ex- terms of whether the Department of Commerce or the port of dual-use technologies and keeping lists updated. Department of State has jurisdiction over export controls. Like COCOM, the WA faces continuously changing mili- Compliance costs were always an issue within COCOM, tary technology. If the WA is to have any effect, “…it and when they became too high relative to the regulatory must at a minimum, identify a set of specific technologies benefits, the system worked poorly. Changing technology, and particular weapons that must not be proliferated.”40 which we discuss below, changes the number of firms that Furthermore, adjusting to changing technologies is made may be able to supply militarily relevant goods and ser- more difficult by shifting relationships between civil and vices, changing both regulatory benefits and compliance military technologies, spin-in rather than spin-off, and in- costs. Traditional approaches to export control may now creasing international economic integration and globaliza- be infeasible. Michael Moodie argues that non-traditional tion. approaches to control will be needed, “…broadening its scope, increasing its flexibility, simplifying its methods, and There have been long cycles in the relationship between enhancing the speed of its accomplishments.”37 This may military and civil technology. After World War II, military involve a shift from controlling supply, as such, to moni- technology was in advance of civilian technology; thus, toring the end use and application of the technology in the military was familiar with the relevant applications of order to stop diversion to military purposes. a technology before it was widely adopted by civilians. In certain cases, the military actually restricted the civilian Supplier groups usually require unanimity to operate adoption of technology in ways that reduced the potential (e.g., to define and change lists, admit new members, etc.). threat. For example, the military restricted the public ad- This is almost certainly unavoidable, but raises problems aptation of the global positioning system (GPS) resolu- of maintaining speed and flexibility. States may also have tion available to civilians, because it controlled all of the incentives to join clubs in order to exploit the unanimity satellites.41 Now that civilian technology is often in the rule by exercising their veto power to inhibit smooth op- lead, the military may find it better to adopt an operating eration of the organization. For instance, some suggested civilian technology, as it did with the iridium global satel- that this was the United Kingdom’s main motivation in lite-communications system. Colonel Robert Weber, pro- joining the EU, (the European Economic Community, or gram manager for the Defense Information Systems ECC, as it was then known). One cannot assume that the network, a unit of the U.S. DOD, stated, “[t]he DOD used desire to join a club signals agreement with its objectives. to be a leader in satellite communications. Now everyone The slow progress of the WA has raised similar suspicions has caught on to space.... We’ll take advantage of [the about the motives of some of its members. Unanimity private sector’s] economies of scale.” 42 Consequences requirements also mean that change will inevitably be in- must be addressed when the private sector finds it un-

86 The Nonproliferation Review/Summer 2001 RON SMITH & BERNARD UDIS

profitable to provide technology, as was the case with the been crucial to the diffusion of technologies. Applications Iridium satellite system. such as word-processing and spreadsheets for personal There are cases where traditional military technologies computers and electronic mail on the internet were not (e.g., encryption and simulation) have such important predicted in advance by those who understood the tech- commercial applications that private firms are willing to nology. invest heavily in research and development (R&D) to These general changes in technology give rise to three duplicate the effective knowledge that the military will not particular sources of military apprehension. The first con- release. When there is a large and obvious profit poten- cerns traditional military products, for example, advanced tial, as with cryptography, private R&D may dwarf that fighters or precision guided weapons. Competitive pres- of the military. Unhampered by bureaucratic and security sure in the arms export market means that firms are in- restrictions, private R&D may also be more flexible, more creasingly selling state-of-the-art equipment. In some cases, innovative, and better organized. In addition, attempts by the recipients may not have the skills in organization or the military to control the technology are likely to con- military doctrine to use these complex systems effectively; front major difficulties. Traditionally, dual-use technolo- in other cases, the systems may be relatively autonomous gies were process technologies, such as the machine tools and simple to use. Argentina, for instance, was unable to that Toshiba sold to the Soviet Union, which were used adjust its air force tactics (matching altitude of attack with to improve submarine propellers. Now, however, they are bomb-fusing) or its army tactics rapidly against the United often product technologies directly useful in combat. The Kingdom during the Falklands/Malvinas War, but it could product may be types of information technology (IT), such use an Exocet to destroy a British warship.44 as: communications systems like fiber-optic networks; ci- The second source of concern is that competitive pres- vilian sub-components that can be assembled into weap- sures are forcing the arms industry to transfer technolo- ons; or civilian products, like remotely piloted vehicles gies as part of the offset requirement for arms deals, thus designed for crop-spraying, but capable of being used as diffusing military technologies. Unfortunately for the re- delivery or reconnaissance vehicles. cipients, they often do not have the broader scientific and The military must constantly consider future technolo- industrial infrastructure to absorb the technology in a way gies and how they may change combat; but usually it fo- that allows them to develop their own systems.45 Tradi- cuses on a few critical military-relevant technologies. It is tional military systems are highly specialized and complex, difficult to forecast the tortuous path that must be fol- and even advanced industrial countries have difficulty with lowed in moving from a technology to a weapons sys- the system integration problems involved. In many cases, tem, then to an appropriate doctrine for its use, and finally technology transfer creates dependence, rather than in- to a change in military organization to deploy the new dependence. system effectively. Now it is more difficult than ever, as The third source of concern centers around the inno- there is a large range of generic technologies that have the vative military applications of generic civilian technolo- capability to revolutionize military affairs. Some of these gies; i.e., the military equivalent of the “killer applications.” are familiar but changing rapidly, like biotechnology and This is the most speculative, but potentially the most dan- IT; others are relatively new, like smart materials, nano- gerous concern: “[o]ne has visions of an Indian software technology, and high temperature super-conductors. These engineer developing a new algorithm that can jam com- technologies are being developed with private rather than mand and control broadcasts, selling it to Iran, which then public funds, often within transnational companies where uses it to bring the U.S. Navy to a halt in the Straight of technology flows globally. Therefore, the technology is Hormuz.”46 It is now common for U.S. software engi- also perceived as central to economic competitiveness and neers to work on a program during the day, beam it by export success. By the time the military utility of such satellite to India, Israel, or Russia, where another team technologies has been identified, significant commercial works on it during the U.S. night and passes it back the interests may have already developed, making controls next morning. Controlling such technology transfer raises more difficult to apply. Even where there are clear tech- obvious difficulties for traditional export control mecha- nological trajectories, as with Moore’s Law, the applica- nisms. Predicting such “killer applications” poses even tions of the technology are unpredictable.43 So-called greater problems for traditional intelligence communities, “killer applications,” made possible by Moore’s Law, have

The Nonproliferation Review/Summer 2001 87 RON SMITH & BERNARD UDIS and the threat from these new technologies will probably mal composition” is important. In principle, the WA is open be taken seriously only when they are unexpectedly used to any country that is a producer or exporter of arms or by a small or revisionist state to change the balance of items subject to control; adheres to the NPT, the BTWC, military power. Having observed the Gulf War and the and the CWC; and maintains and applies effective national conflict that occurred in Kosovo, revisionist states might export controls. The WA, with 33 members as of June realize that they should not attempt to compete on tradi- 2001, is a fairly large group that lacks the coherence that tional battlefields, where the United States dominates. was provided to COCOM by the shared perception of the Instead, they are likely to turn to asymmetric means of Soviet threat. Building a coherent vision of shared inter- warfare, using generic capabilities that have been widely ests is important. The WA is a supplier group and will diffused, rather than traditional military technologies mo- inevitably lack legitimacy on the demand side. According nopolized by the larger military powers. to some, supplier groups are basically cartels with a common interest not merely in preventing the spread of de- THE WASSENAAR ARRANGEMENT stabilizing arms and technologies, but in restricting supply to raise prices. Such groups are seen by the “have-nots” With the end of the Cold War, COCOM was no longer as an attempt by the “haves” to maintain their monopoly viable or legitimate. At a high-level meeting in The Hague positions in arms and technology. To increase the legiti- on November 16, 1993, members decided to phase out macy of the WA, important outreach measures to inte- the organization by March 31, 1994. After considerable grate recipient states into the agreement are conducted. discussion, the WA was formed, named after the town in Outreach efforts may also include participation in regional the Netherlands where it was provisionally established in restraint discussions aimed at establishing demand-side 1995. The WA was formally established in Vienna, Aus- agreements to stop the diffusion of particular systems or tria, in July 1996. It differs from COCOM in its member- technologies to a region. ship and method. Russia and other members of the former Warsaw Pact participate, although China does not; and At the June 1997 meeting of the General Working Group transparency is a key feature, rather than the secretive- of the WA, the United States tabled a paper entitled, ness that surrounds a control regime. The principal goal “Emerging Weapons of the 21st Century.” The paper of the WA is to gather information, which reportedly re- identified five areas where the United States expects the veals potentially destabilizing accumulations of weapons development and improvement of advanced weapons ca- or militarily useful dual-use technologies. pabilities within the next 15 to 20 years. These areas included: WMD; long-range precision weapons; recon- Participating states agree to maintain national controls naissance, surveillance, and target acquisition systems; on the transfer of items specified in the dual-use and mu- counters to precision strike capabilities; and information nitions lists; notify other members of aggregate transfers warfare. The United States argued that these issues were of munitions to non-participating states; and notify other of concern to the WA because of the threat of prolifera- members of individual cases where licenses to transfer tion of such capabilities through exports, diversion, or the dual-use items are denied. In principle, unlike COCOM, upgrading of existing weapons. U.S. representatives the WA is not directed at any state or group of states; in pointed to the danger of increasing the combat capability practice, it is directed at countries whose behavior is a of states that might deploy them in an irresponsible or cause for serious concern, particularly Iran, Iraq, Libya, destabilizing way. and North Korea. The first operational plenary session was held in Vienna in December 1997. The WA is still a The traditional method of handling technology change young arrangement, and while this immaturity contributes is through a process of list review, where old items are to its current ineffectiveness, it may make it sufficiently removed and new items added—a process that tends to pliable to be shaped in ways that could allow it to evolve lag behind very rapid changes in technology. Within into a more effective organization in the future. COCOM, list review was fraught with difficulties, and lists grew quickly out of date, sometimes due to older tech- Within the WA, the “Big 6” (France, Germany, Italy, nologies that were widely available. Of particular danger Russia, the United Kingdom, and the United States) have are precursor technologies, which present the possibility agreed informally to meet for more intensive consultations of vaulting over the traditional stages in the development and more intrusive information sharing. This may poten- of militarily useful products and technologies. Presently, tially lead to a subset within the WA. The issue of “opti-

88 The Nonproliferation Review/Summer 2001 RON SMITH & BERNARD UDIS

the WA lacks the means to identify such developments; decisionmaking at the national level would help to trans- its munitions list includes only the seven categories of the mit insights and increase the pace of decisions in the vari- U.N. arms transfer register.47 This excludes many muni- ous arrangements. At an institutional level, the WA must tions, particularly retrofit equipment, which can substan- develop an effective monitoring system, including its own tially upgrade the military capability of old tanks, combat IT system to track movements of weapons and sensitive aircraft, or warships. Extending the list has been the sub- technologies, based on information provided by national ject of discussion within the WA recently. Sometimes, participants. This would also increase the autonomy of though, it is also necessary to pull items off of the list or the WA. further define covered technologies that have wide civil- Alternative approaches do exist. Wolfgang Reinicke ian use. For example, at the December 1998 meeting, the proposed an innovative arms export control system that control lists were changed to reduce coverage of civil tele- operates on a self-regulatory basis at the level of the indi- communications and to update controls on encryption. vidual firm, a model derived from financial self-regula- Different states are subject to different domestic, po- tory organizations (SROs).49 This would be very different litical, and bureaucratic constraints, as well as varying in- from the traditional WA framework, and it is not clear ternational perceptions. Several states have expressed fears whether a multilateral system of this sort would be fea- that the United States, traditionally the dominant supplier, sible. Relative to finance, far more firms would be involved, attempts to use controls to further its own interests and to and there is no natural definition of the market, since so maintain its monopoly. This was a recurring problem within many technologies can be used for military purposes. In COCOM as well. Despite the fears of U.S. exploitation addition, the activities subject to control are not as well of its powerful international position, the United States is defined as in finance, and arms sales are far more politi- the natural leader, if only because it exports the most arms. cally sensitive than financial transactions. On the logisti- However, the United States, partly because of the consti- cal side, the monitoring requirements seem to exceed tutionally imposed separation of government powers, has current IT and reporting capabilities. Finally, there is no difficulty providing consistent, coherent leadership over experience to draw upon with respect to international long periods. Instead, there is a process of sequential at- SROs, even in finance.50 tention to different goals, as different issues become sa- In drawing lessons from other arrangements, it is per- lient in Washington. This myopia has weakened U.S. haps instructive to observe the progress made by the leadership. Myopia is not just a U.S. problem; France has Anglo-French proposal for a “Code of Conduct for Arms often appeared myopic due to its focus on short-term eco- Exports,” agreed to by EU Ministers in May 1998. The nomic self-interest at the expense of longer-term common code outlines eight guidelines to govern the process by interest. which states give licenses; establishes consultation mecha- Another institutional feature that shapes the effective- nisms between states; and requires all EU countries to ness of the arrangement is that, for bureaucratic reasons, submit annual reports on arms exports. The Anglo-French the United States, unlike other nations, tends to separate proposal originally faced opposition from other EU mem- its delegation membership by regime. Rarely do the same bers—smaller countries were concerned that the controls persons participate in all of the various arrangements. As were not strict enough and that consultations were only a consequence, if items arise at MTCR or AG meetings bilateral (between the state that first denied a license and that are relevant to the WA, the U.S. representative has the state considering supplying) rather than multilateral. difficulty in responding on the spot in a meaningful way.48 In the negotiations, the conflicting interests between the Conversely, the French are quite centralized; if the cru- main exporters, Britain and France, and the other EU cial decisionmaker has not made a decision, the represen- countries were apparent. tatives at all of the arrangements must stall the overall decision. GAME THEORY DETERMINANTS Technology, however, does not respect these artificial While we cannot make firm predictions about whether boundaries between regimes. The creation of a compre- the WA will develop into an effective organization, we can hensive regime combining all of the existing control and identify certain factors drawn from game theory that could information sharing systems would not be sensible. Nev- be determinant. ertheless, systematic cross-fertilization and coherent

The Nonproliferation Review/Summer 2001 89 RON SMITH & BERNARD UDIS

The first concerns the structure of payoffs, or incen- slow. This would require suppliers to delegate authority tives. The long-term payoff is the security benefit of a to the WA, making it a much more autonomous body. reduction in proliferation. To the extent that the partici- Fourth, the degree of credibility in the system is essen- pants share the perception of a serious threat, they will tial. The continuity of policies and the certainty of conse- work together and solve the technical problems of con- quences must be clear. Each member must know that trol. This is a lesson from COCOM: the Europeans coop- defection will be detected (monitoring) and punished (cred- erated with the United States best when they perceived a ibility), while compliance will be recognized and rewarded. serious Soviet threat. However, it may take more than At present, the WA has little credibility. the Indian and Pakistani nuclear tests to persuade the participants that new technologies pose a serious threat. Rather, Finally, the degree of leadership cannot be ignored. In it may require the use of a “killer application” by a revi- the case of the WA, concerted commitment by the major sionist power. The short-term payoffs, on the other hand, arms producers would be required—particularly the United involve the economic consequences of a sale. At present, States, the United Kingdom, and France—in order to economics reward non-compliance, as there are profits change the structure of payoffs (putting more weight on to be made from (even potentially destabilizing) sales. long-term security consequences); develop an effective There may be scope to design rewards for compliance as monitoring capability; and infuse the arrangement with was done with the NPT, soliciting Russian adherence to more credibility and autonomy. Without this commitment, the MTCR, and the denuclearization of the former So- which currently does not exist, it is unlikely that Russia viet Republics. and the smaller producers will cooperate, or that the WA will be effective. A second factor is the degree of myopia that states possess, or the discount rate. Myopia may arise for a variety of reasons, since different motivations exist for different countries, as was witnessed in the cases of France and 1 The opinions in this paper are those of the authors and should not be the United States. For the WA to prove successful, the attributed to the U.S. government. Smith is grateful to the Department of participants must accept short-term economic and politi- Economics, University of Colorado at Boulder for its hospitality during 1997- cal costs in return for long-term security benefits. The 1998, when most of the work on this paper was done, and to the ESRC for financial support under research grant R00235685 on models of the arms trade. relative weight given to the present and the future is thus We are also grateful to Mary Ann Casey, Yongmin Chen, Saadet Deger, Paul crucial. An imminent threat, of course, reduces myopia Levine, and Somnath Sen for comments on an earlier draft. 2 Paul Levine and Ron Smith, “The arms trade game: from laissez-faire to a with regard to security consequences. For instance, the common defence policy,” Oxford Economic Papers 52 (2000), pp. 357-380. Iraqi invasion of Kuwait, and the subsequent discovery 3 For a history of COCOM, see Michael Mastanduno, Economic Contain- of a covert nuclear weapons program in Iraq, increased ment: COCOM and the Politics of East West Trade (Ithaca, NY: Cornell Uni- versity Press, 1992). security concerns among suppliers and motivated them 4 Andrew J. Pierre, ed., Cascade of Arms (Washington, D.C.: Brookings for the to enhance control. The effect wore off, however. Re- World Peace Foundation, 1997), p. 398. viewing the lessons from the conventional arms transfer 5 William Keller and Janne E. Nolan, “The Arms Trade: Business as Usual,” Foreign Policy, No.109 (Winter 1997-1998), p. 123. (CAT) control negotiations, Janne Nolan commented on 6 Useful references are Pierre, ed., Cascade of Arms; Robert E. Harkavy and the danger of “…a permanent ascendance of short-term Stephanie G. Neuman, eds., “The Arms Trade: Problems and Prospects in the 51 Post-Cold War World,” The Annals of the American Academy of Political and over long-term interests.” Social Science No. 535 (September 1994); David Mussington, Understanding Third, the monitoring capability of the system (or lack Contemporary International Arms Transfers, Adelphi Paper 291 (London: International Institute of Strategic Studies, 1994); Peter van Ham, “Managing thereof) is significant. While the main source of monitor- Non-Proliferation Regimes in the 1990s,” Chatham House Papers, (London: ing involves national export control systems and intelli- Printer Publishers for the Royal Institute of International Affairs, 1994); and RAND, Arms Proliferation Policy: Support to the Presidential Advisory Board gence sources, the WA could increase the degree of (Santa Monica, CA: RAND, 1994). transparency by pooling and tracking information. This 7 These P-5 guidelines are discussed in Pierre, ed., Cascade of Arms, pp. 377- would also serve to increase shared perceptions of poten- 380. 8 Colin S. Gray, House of Cards: Why Arms Control Must Fail (Ithaca, NY: tially threatening developments in technology, intentions, Cornell University Press, 1992). or capabilities. To do this effectively will require better 9 Pierre, ed., Cascade of Arms, p. 4. 10 IT capability and more flexible procedures to identify what The theory of the decision as to when a country becomes a producer of arms is examined in Paul Levine, Fotis Mouzakis, and Ron Smith, “Arms Export should be monitored. Traditional list review procedures Controls and Emerging Domestic Producers,” Defence and Peace Economics to determine what will be monitored are likely to be too 11 (September 2000), pp. 505-531; and Paul Levine and Ron Smith, “Arms

90 The Nonproliferation Review/Summer 2001 RON SMITH & BERNARD UDIS

Export Controls and Proliferation,” Journal of Conflict Resolution 44 (Decem- plexity (Princton, NJ: Princeton University Press, 1997). ber 2000), pp. 885-895. 26 Kenneth W. Abbott and Duncan Snidal, “Why States Act Through Formal 11 For a review of the trade of light weapons, see Jeffrey Boutwell, Michael T. International Organizations,” Journal of Conflict Resolution 42 (February Klare, and Laura W. Reed, eds., Lethal Commerce: The Global Trade in Small 1998), p. 9. Arms and Light Weapons (Cambridge, MA: American Academy of Arts and 27 The dynamics of the international negotiations involved are discussed in Sciences, 1995). For a discussion on the control of light weapons, see Owen Fen Osler Hampson and Michael Hart, Multilateral Negotiations: Lessons Greene, Tackling Light Weapon Proliferation: Issues and Priorities for the from Arms Control, Trade and the Environment (Baltimore, MD: The Johns EU (London: Saferworld, 1997). Hopkins University Press, 1995). 12 See Mastanduno, Economic Containment. 28 A framework for examining the relationship between the various patches is 13 A classic statement of arms control theory can be found in Thomas C. given in Robert E. Harkavy, “Harmonizing Policies across Arms Control Do- Schelling and Morton H. Halperin, Strategy and Arms Control (New York: mains: Dilemmas and Contradictions,” in R. E. Harkavy and E. Kolodziej, eds., Twentieth Century Fund, 1962). Gray provides a fundamental critique of the American Security Policy and Policy Making: The Dilemmas of Using and whole activity. See Gray, House of Cards. Controlling Military Force (Lexington, MA: Lexington Books, 1980), pp. 129- 14 The economics and security trade-offs and their implications for control 147. have been analyzed in the context of formal models. See Paul Levine and Ron 29 For a non-technical discussion, see Stuart Kauffman, At Home in the Uni- Smith, “The Arms Trade and Arms Control,” Economic Journal 105 (March verse (Oxford: Oxford University Press, 1995), Ch. 11. 1995), pp. 471-484; and Paul Levine and Ron Smith, “The Arms Trade,” Eco- 30 Pierre, ed., Cascade of Arms, p. 428. nomic Policy, No. 25 (October 1997), pp. 335-370. 31 There are of course cases where they overlap, and some arms transfer 15 The early development of EU policy is reviewed in Paul Cornish, “The Arms negotiations with the Soviet Union or Russia involve very traditional ele- Trade and Europe,” Chatham House Papers (London: the Royal Institute of ments. Two examples are: (1) the U.S.-Soviet negotiations on conventional International Affairs, 1995). arms transfer (CAT) control, discussed in Janne E. Nolan, “The U.S.-Soviet 16 These agreements are described in Stockholm International Peace Research Conventional Arms Transfer Negotiations,” in Alexander L. George, Philip J. Institute (SIPRI), SIPRI Yearbook, Armaments, Disarmament and Interna- Farley and Alexander Dallin, eds., U.S.-Soviet Security Cooperation: Achieve- tional Security (Oxford: Oxford University Press, 1999), Annex A. ments, Failures, Lessons (Oxford: Oxford University Press, 1988), pp. 510-523; 17 The term “regime” has a specialized meaning in international relations theory; and (2) the U.S.-Russian negotiations on the MTCR, described in Pikayev et for instance, see Mastanduno, Economic Containment, p. 5. He states, “By al., Russia, the US and the Missile Technology Control Regime. ‘regime,’ I mean a set of rules, norms and expectations around which actors’ 32 Todd Sandler and Keith Hartley, The Economics of Defense (Cambridge: expectations converge in a given issue area.” It is a subject of dispute as to Cambridge University Press, 1995). whether these agreements are regimes in this sense, so we have avoided the 33 See Kauffman, At Home in the Universe. See also Gray, House of Cards. term. For a discussion on what an arms control regime might involve, see 34 Michael Moodie, “Constraining Conventional Arms Transfers,” in Harkavy Pierre, ed., Cascade of Arms, Ch. 15. For Levine and Smith, a regime is used to and Neuman, eds., The Arms Trade: Problems and Prospects, p. 136. denote particular combinations of forms of coordination. See Levine and Smith, 35 For discussions on the use of black and gray markets to acquire arms, see “The arms trade game.” Aaron Karp, “The Rise of Black and Gray Markets,” in Harkavy and Neuman, 18 Alexander A. Pikayev, Leonard S. Spector, Elina V. Kirichenko, and Ryan eds., The Arms Trade: Problems and Prospects, pp. 175-189; and Michael T. Gibson, Russia, the US and the Missile Technology Control Regime, Adelphi Klare, “The Subterranean Arms Trade: Black-Market Sales, Covert Operations Paper 317 (London: International Institute for Strategic Studies, 1998). and Ethnic Warfare,” in Pierre, ed., Cascade of Arms, pp. 43-71 19 For a discussion of the implications of this, see Wolfgang H. Reinicke, 36 See Reinicke, “Cooperative Security.” “Cooperative Security and the Political Economy of Nonproliferation,” in Janne 37 Moodie, in Harkavy and Neuman, eds., The Arms Trade, p. 132. Nolan, ed., Global Engagement: Cooperation in the 21st Century (Washing- 38 See Axelrod, The Complexity of Cooperation, Ch. 3; Robert D. Cooter, “Law ton, D.C.: Brookings Institution, 1994), pp. 175-234. from Order: Economic Development and the Jurisprudence of Social Norms,” 20 See Wade Boese, “US Buys Moldovan Aircraft to Prevent Acquisition by in Mancur Olson and Satu Kahkonen, eds., A Not So Dismal Science (Oxford: Iran,” Arms Control Today 27 (1997), p. 28. Oxford University Press, 2000), Ch. 9. 21 A formal analysis of such measures is given in Philippe Jehiel, Benny 39 For a discussion of the treaty, see Cassady B. Craft, “An Analysis of the Moldovanu, and Ennio Stacchetti, “How (Not) to Sell Nuclear Weapons,” Washington Naval Agreements and the Economic Provisions of Arms Con- American Economic Review 86 (September 1996), pp. 814-829. trol Theory,” Defence and Peace Economics 11, No. 2 (2000), pp. 127-148. For 22 The magnitude of such economic benefits to the supplier states is a matter military technology and the arms trade, see W. Seth Carus, “Military Technol- of dispute. Commercial confidence requires that little of the detail is disclosed, ogy and the Arms Trade: Changes and Their Impact,” in Harkavy and Neuman, and the contracts are very complex, involving subsidized R&D, soft credit, eds., The Arms Trade, pp. 163-174. and offsets. Firms clearly make profits, for they would not accept the contracts 40 William Keller and Janne E. Nolan, “The Arms Trade: Business as Usual,” otherwise; but these profits may come from supplier subsidies and not recipi- Foreign Policy, No. 109 (Winter 1997-1998), pp. 123-124. ent payments. For a discussion of financial aspects of the arms trade, see Joel 41 See Irving Lachow, “The GPS Dilemma: Balancing Military Risks and Eco- L. Johnson, “Financing the Arms Trade,” in Harkavy and Neuman, eds., The nomic Benefits,” International Security 20 (Summer 1995), pp. 126-148. Arms Trade, pp. 110-121. Trends in the mid-1990s are reviewed in George 42 Wall Street Journal, January 26, 1998, p. B4. Anayiotis and Nancy Happe, “Recent Trends and the Financing of the Arms 43 Moore’s law: that the number of components on a chip or computer power Trade,” Peace Economics, Peace Science and Public Policy 4, No. 4 (1997), per dollar doubles every 18 months. Moore proposed this concept in 1965, pp. 1-30. The costs and benefits of offsets are discussed in Bernard Udis, and it still seems to hold, though it must hit physical limits eventually. Offsets in Defense Trade: Costs and Benefits, unpublished NATO Research 44 For a discussion about the failure of the MTCR to control cruise missiles, Report, 1994. See also Bernard Udis and Keith E Maskus, “Offsets as Indus- see Dennis M. Gormley, “Hedging Against the Cruise Missile Threat,” Sur- trial Policy: Lessons from Aerospace,” Defense Economics 2 (1991), pp. 151- vival 40 (Spring 1998), pp. 92-110. 164. 45 See Stephen Martin, ed., The Economics of Offsets: Defense Procurement 23 Such collective action problems are discussed in Todd Sandler, Collective and Countertrade (London: Harwood, 1996); and Joanna Spear, “The Role of Action (Ann Arbor, MI: University of Michigan Press, 1992); and Todd Sandler, Offsets in the International Arms Trade,” paper presented at the ISA Annual “Arms Trade, Arms Control and Security: Collective Action Issues,” Defence Meeting, Toronto, April 1997. and Peace Economics 11 (September 2000), pp. 533-548. 46 For an example, see David Silverberg, “Global Trends in Military Production 24 See Robert Axelrod, The Evolution of Cooperation (New York: Basic Books, and Conversion,” in Harkavy and Neuman, eds., The Arms Trade, p. 126. 1985). 47 Battle tanks, armored combat vehicles, large caliber artillery systems, com- 25 Robert Axelrod, The Complexity of Cooperation, Princeton Studies in Com- bat aircraft, attack helicopters, warships, and missiles/missile launchers.

The Nonproliferation Review/Summer 2001 91 RON SMITH & BERNARD UDIS

48 For a discussion on the limitations of the U.S. export control process, see indicated how a more fully developed version of his model might work to RAND, Arms Proliferation Policy, Chapter 7. For a discussion on the impact correct these deficiencies. See Wolfgang H. Reinicke, Global Public Policy: of bureaucratic disputes on the CAT, see Nolan, “U.S.-Soviet Conventional Governing Without Government (Washington, D.C.: Brookings Institution Arms Transfer Negotiations.” Press, 1998), pp. 173-217. 49 See Reinicke, “Cooperative Security.” 51 See Nolan, “U.S.-Soviet Conventional Arms Transfer Negotiations,” pp. 50 We should note, however, that in a more recent work, Reinicke has pre- 510-523. sented a devastating critique of the structure and functioning of the WA and

92 The Nonproliferation Review/Summer 2001